GEE (CCDC-2) - Land cover classification analysis based on the CCDC coefficient of the study area obtained

Introduction

According to the CCDC land classification image of the study area and the selected training sample points, the CCDC segmentation image is then obtained and loaded. This is equivalent to the second part of CCDC. The first part is to obtain long-term multi-band remote sensing. , The second part is to obtain the classification results of the specified area based on your own research area and training sample points. This tutorial explains in detail what should be paid attention to in the second part and how to add your own training sample points, as well as the detailed tutorial of CCDCapp.

CCDC APP

https://code.earthengine.google.com/?scriptPath=users%2Fparevalo_bu%2Fgee-ccdc-tools%3AAPPS%2Flandcover_app

In this interface, we can monitor points one by one on the left side of the APP. The main function is that we can click on any point and see its changes and mutations.

 

Research data import 

For the analysis of regional area, we need to use the right panel. First, we download the file in CCDC-1.

GEE (ccdc-1) - Use Landsat series images to obtain multi-band images within the study area (including multi-bands already defined in ccdc) - CSDN blog is imported into the folder path above. Note that this is a single-band image. The default here is an image collection, and there are also vector collection options to choose from, so we need to modify it. Of course, there are also three time format options, the first is day, the second is year, and the third is The third is the time in milliseconds. You can choose it according to your needs. Generally, the default is enough.

 Classification parameter selection

Here we need to select a classifier. Here we provide three classifiers: random forest, support vector machine and Euclidean distance. Then during the model building process, we can choose many variables to participate. You can choose according to your own situation.

 

Export scope settings

Here we can choose a variety of methods to draw and specify the research area. If we have our own research area, we can select feature collection and then export the results according to the scope of our own research area. At the same time, we can also choose other methods, single tile and multi-tile superposition to select the output range.

 For a single tile, you need to enter the row, column number and state. This grid is introduced in the previous blog section.

 Select points of interest

Multiple tile options

 Select by country border

Choose according to ecological zone

 

Upload the research path of your assets here. When you upload it here, your research area will be automatically displayed on the map.

 

Training data upload

Here we can see the final upload of training data. The first is the training data strategy. Here we generally select the specified range of the research area, then upload our training sample point path, then select our classification attribute name, and then select Specify The year of the training sample point. 

 

Here we can print and see what the original sample point data provided in the article includes, so that we can upload the correct land classification sample points. Here it contains all the bands and selected data, and of course attribute information. Remember to enter landcover as The attributes of the classification, that is, the attribute names in the land classification

var ss = ee.FeatureCollection('users/bullocke/ccdc/public/EastAfrica_Training_Example_Data')
 print(ss.limit(10))
 Map.addLayer(ss,{},'samples')

original sample point 

 

Here we filter the sample points in our own research area and put them in. The code of the training sample points is similar to the code below, then export it to our own Google assets, and then fill in the path space above.

//在GEE中筛选好的样本点进行合并
var newfc = corp
                    .merge(grass)
                    .merge(building)
                    .merge(bare)
                    .merge(coniferious)
                    .merge(broadleaf)
                    .merge(mining);
//var training = ee.FeatureCollection(training)
print("newfc",newfc)

//CCDC SAMPLES/
// 添加指定的属性信息，这里这么做的主要目的就是能让软件读取你的信息，然后顺利的进行分类分割影像的监测。
var trainingData  = newfc.map(function(feat) {
  return feat.set('lc_numeric',feat.get('landcover'))})
  .remap([0,1,2,3,4,5,6],[1,2,3,4,5,6,7],'lc_numeric')
print(trainingData.limit(10))
// 添加时间属性
var trainingData  = trainingData.map(function(feat) {
  return feat.set('year',2022)})
print(trainingData.limit(10))
  
Export.table.toAsset({collection:trainingData, description:'hexinkuangqu'})

 There will also be a legend on the display showing whether to use glance land classification to construct your classification results.

GlanCE  data

NASA produces Earth System Data Records for Research in the Environment (  MEaSUREs  ) Global Land Cover Mapping and Estimation (  GLanCE  ) annual 30 meter (m) version 1 data products ( https://lpdaac.usgs.gov/products/glance30v001/ ) available includes global land cover and land cover change data from Landsat 5 Thematic Mapper (TM), Landsat 7 Enhanced Thematic Mapper Plus (ETM+) and Landsat 8 Operational Land Imager (OLI). These maps provide user communities with land cover types, land cover changes, indicators characterizing the greenness and seasonality of each pixel, and the magnitude of change. GLanCE data products will be provided using a set of seven continental grids using parameterized Lambert azimuthal equal-area projections to minimize distortion for each continent. Currently available in North America and continental Europe. This dataset can be used for a wide range of applications, including ecosystem, climate, and hydrological modelling; monitoring terrestrial ecosystem responses to climate change; carbon accounting; and land management.

The GLanCE data product provides seven layers: land cover category , date of estimated yearly change, integer identifier for the previous year category, median and magnitude of the enhanced vegetation index (EVI2) for the current year, EVI2 change rate, and median EVI2 change from the previous year Changes to this year. Each particle also provides a low-resolution view of the EVI2 amplitude.

Characteristics

Collection and Granule

Collection

Characteristic	Description
Collection	MEaSUREs GLanCE
DOI	10.5067/MEaSUREs/GLanCE/GLanCE30.001
File Size	~250 MB
Temporal Resolution	Yearly
Temporal Extent	2001-07-01 to 2019-07-01
Spatial Extent	Global
Coordinate System	Lambert Azimuthal Equal Area
Datum	World Geodetic System (WGS84)
File Format	GeoTIFF
Geographic Dimensions	Global

Granule

Characteristic	Description
Number of Science Dataset (SDS) Layers	7
Columns/Rows	5000 x 5000
Pixel Size	30 m

Layers

SDS Name	Description	Units	Data Type	Fill Value	No Data Value	Valid Range	Scale Factor
LC	Integer identifier for class in the current year	Class	8-bit unsigned integer	255	N/A	1 to 7	N/A
ChgDate	Estimated day of year of change; N/A if no change	Day	16-bit unsigned integer	32767	N/A	1 to 365	N/A
PrevClass